It has been incredibly humid around these parts over the last week, and there seems to be something about these dog days that makes you leave the fridge or freezer door open by mistake. [pnjensen] found this happening all too often to the family chill chest, with the predictable accretion of frost on the coils as the water vapor condensed out of the entrained humid air and froze. The WiFi-enabled fridge alarm he built to fight this is a pretty neat hack with lots of potential for expansion.
Based on a Sparkfun ESP8266 Thing and home-brew door sensors built from copper tape, the alarm is rigged to sound after 120 seconds of the door being open. From the description it seems like the on-board buzzer provides a periodic reminder pip while the door is open before going into constant alarm and sending an SMS message or email; that’s a nice touch, and having the local alarm in addition to the text or email is good practice. As a bonus, [pjensen] also gets a log of each opening and closing of the fridge and freezer. As for expansion, the I2C header is just waiting for more sensors to be added, and the built-in LiPo charger would provide redundancy in a power failure.
If frost buildup is less a problem for you than midnight snack runs causing another kind of buildup, you might want to check out this willpower-enhancing IoT fridge alarm.
Now that I think about it, how come fridge doors don’t have spring returns on them like screen doors?
I recall that when our current fridge was delivered, the guy who set it up made sure that it wasn’t perfectly level. He put a definite backwards tilt on it using the leveling legs, which annoyed the hell out of me and my wife, OCD-ers that we are. But he explained that it was to not only make sure the doors would self-close, but that they’d have enough velocity to seal effectively. Seems to work, even if it looks a little out of plumb.
Yeah, the Jello (TM) in your Jello (TM) mold is not level!
A lot of them do, but not in the forms of springs, a common failure point on some of these problems occur when the plastic or Teflon spacer the door hing rests on degrades, its angled to “encourage” the door to close, this requires the fridge to be mostly level, although tilting it back a little would help, they try to keep that to a minimum to keep round things from wandering off when stored in the fridge.
And btw, velocity isnt required to seal the door Dan. there is a magnetic strip inside the gasket to help make the seal. the biggest problem I have with that is when I have to change a seal and the new one comes all twisted up in a box and I have to use a heat gun to relax it and wipe it down with a cold rag after, in a commercial kitchen they usually have a large sink i can fill with hot water, THAT’s Ideal. but for the home, this would work, a bit over kill, but nicely done.
I would love to see this become a more sensitive light switch sensor instead of the bump switch thats currently used, but im afraid it might get a false reading if it gets wet.
I think most Lithium polymer batteries aren’t good in freezing temperatures. Sanyo eneloop NiMH batteries are rated for use well below freezing.
Lithium are fine in freezing, just do go below -40ish. There Discharge rate will be slower at lower temps, but for this purpose it should be fine.
Actually, as the true owner of this IOT fridge, the device doesn’t live inside the compartment – but outside…The wifi radio inside a metal box doesn’t work so well… In addition, pnjensen choose the foil contact idea SPECIFICALLY because when frost builds up on the gasket area, the door APPEARS to close when it is almost-shut-but=isn’t. The foil contacts make a *very sensitive* distance sensor – so that they indicate ‘door open’ with the slightest door-ajar situation. Using a ‘light sensor’ would not have worked in our situation because the ‘light switch’ trips well before the door actually seals. The next extension to this will be a temp sensor on the evaporator coil to confirm the defrost cycle is occurring regularly. The cheapo electrolytic caps in the OEM defrost timers dry out quickly in the dry environment, causing the defrost timer to never trip.
Things are just too easy. The old doors had a latch and it took all of one hand to open. Just a little training on using a refrigerator is all that is needed. Such petty “convenient” excuses include running water while the fridge door is open and the burner is on waiting for a pan.
At last, someone is using an ESP8266 and it’s not that crappy ESP-01 board!
Instead of a time consuming hack, one could also use the light sensor on this new Finnish IoT product Thingsee One to tell, if the door is open.
The problem with that is that the fridge light turns off well before the door is actually sealed shut. If you try to detect light leaking around the edge of the door, you will get a negligible increase in the amount of light hitting the sensor, as it has to sneak in between the seal and door frame, and then make a 90 degree turn and sneak in between the shelves and the wall. This is probably a three to six inch path, which isn’t really feasible for light to make it all the way in, unless you use external illumination.
All in all, it took me less than 30 seconds to make the sensors, 3 hours to make the board in a box (which can be used for different applications), and a day for code. The code was the slow part for this project.
With things like the June oven being worked on, I’m surprised someone has tried a “smart” fridge project. Might be in Nest’s wheelhouse, actually. Apply standard machine learning to level loads, or apply extra cooling power when someone is coming home from the grocery store, etc. With a variable speed compressor, this could be pretty amazing.